1. Field of the Invention
The present invention relates to a targeted object pursuit system that pursues the movement of a specified targeted object within a plurality of images captured sequentially.
2. Description of the Related Art
Image analysis to pursue the movement of a desired object within sequentially captured images is known. Such image analysis is used in an auto focus function to pursue a moving object. In addition, such image analysis is used by a monitor camera to pursue a specified person.
The image analysis in prior art is explained briefly below. First, a specified area within the frame of an image to be captured is initially designated as a pursuit area. Second, an image of the pursuit area captured at a certain point in time is compared to images of subsequently captured areas adjacent to the pursuit area, according to pattern matching. In pattern matching, multiple likeness values quantifying the similarity between two compared images are calculated, and area corresponding to the highest likeness value is designated as a new pursuit area. Since then, an object included in the pursuit area is pursued by repeating the above operation.
In the above pursuit method, the area corresponding to the highest likeness value is considered to be the area where the targeted object has moved. Accordingly, the image of such an area does not always accord to the image of the pursuit area.
The image analysis is adapted to the auto focus function by which an object moving within a captured image is brought into focus, as described above. In the image analysis, a hand shake may affect the calculation of a high likeness value irrespective of the similarity between two images. In addition, if the contrast between the images of the pursuit area and the compared areas is low, all likeness values may become too low to compare to one another. Consequently, it is problematic that a targeted object cannot be pursued accurately.